Repository navigation
Conversation
Simulated-user trials previously only ran populate_context_post_run on the user agent, so agent/trajectory.json was never written and verifiers could grade an empty trajectory.
|
@gyanu2507 is attempting to deploy a commit to the Harbor Framework Team on Vercel. A member of the Team first needs to authorize it. |
Contributor
|
Enjoy a better diff viewing experience by clicking one of these URLs: |
![devin-ai-integration[bot]](https://avatars.githubusercontent.com/in/811515?s=60&v=4){kind=small}
This comment was marked as resolved.
This comment was marked as resolved.
Sorry, something went wrong.
Simulated-user trials store target agent tokens under metadata["target"]. Job and viewer totals previously ignored that payload.
Author
|
Included |
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Fixes #3971. With a simulated user agent,
_populate_agent_contextonly ranpopulate_context_post_runon the user agent. Most target agents writeagent/trajectory.jsonin that hook, so verifiers graded an empty trajectory and target token/cost accounting was skipped.This change always runs the target agent's post-run hook on a separate
AgentContext, stores non-empty target usage underagent_result.metadata["target"], and still calls the user agent's hook when the shared context is empty.Test plan
PYTHONPATH=src pytest -p no:randomly tests/unit/trial/test_populate_agent_context.py(5 passed)--agent claude-codeproducesagent/trajectory.jsonafter the run